clock menu more-arrow no yes

Filed under:

Projecting the Big Ten's conference race using advanced stats

New, comments

This week on Win Projections Wednesday, we're using the most recent ratings to project how each Big Ten team's conference record might play out.

The Big Ten hasn't exactly won the optics battle this September. Only two teams have survived four weeks without a loss, three have lost twice, and none ranks higher than 12th in the current F/+ rankings.

We can talk about the shifting demographics of the football universe, we can talk about the need to hire better coaches ... we can talk about a lot of things. But commissioner Jim Delany's conference hasn't won many big games on the field in 2014, meaning it hasn't won many PR points. That's the bad news.

The good news, as it were? Big Ten teams are about to start playing each other. And as with previous iterations of the ACC and Big East, when you have a lot of teams in FBS' mushy middle, you often end up with one hell of a conference race.

Now, while we can quibble with the ELITENESS of the conference's top tier, there is still a top tier. The odds are pretty good that we're looking at a Big Ten championship game of Wisconsin vs. the Ohio State-Michigan State winner. But with these three teams only ranking in or near the F/+ teens, and with a glut of teams ranked between 31st and 46th, the second tier is just close enough to make this race a giant mess with a minimal number of upsets.

Each week at Football Study Hall, I post the weekly picks and projections. The F/+ ratings don't do much better than Vegas in picking games, but they've done an excellent job of establishing win probabilities. Heading into Week 4, teams given a 50-70 percent chance of winning a given game were winning 59 percent of the time, while teams given better than 70 percent chances were winning 90 percent of the time. (You can find more detail at the link above.) As more games happen and the data gets a little sharper (and who's really good and really bad becomes clearer), the probabilities are likely to become even more accurate.

What can these potentially conservative win probabilities tell us about the B1G race? Let's take a look.

Win projections

East F/+ Ranking Proj. Conf. Wins
Michigan State 12 5.83
Ohio State 17 5.47
Penn State (1-0) 33 4.83
Michigan 32 4.64
Indiana 39 3.90
Maryland 46 2.94
Rutgers (0-1) 52 2.30
West F/+ Ranking Proj. Conf. Wins
Wisconsin 15 6.26
Nebraska 31 4.92
Iowa 44 4.69
Minnesota 63 3.26
Northwestern 68 2.94
Illinois 75 2.45
Purdue 93 1.56

Poor Rutgers and Maryland were stuck in the wrong division. (Well, it's the right division geographically, but the wrong one competitively.) The East features five of the conference's seven top-40 teams and none of its four 60th-or-worse teams. Rutgers would be the fourth-best team in the West but is dead last in the East. That tamps down expectations, especially after the Scarlet Knights let a potential home win over Penn State slip away at the last second, and after running back Paul James was lost for the season with a knee injury. Plus, the Knights draw Wisconsin (home) and Nebraska (away) from the West.

The projections basically follow the rankings, as you would expect. Penn State has the best schedule of the East's mid-tier teams, and the Nittany Lions have a 100 percent chance of finishing with a win over Rutgers. So while they might not be as strong as some think -- when their NCAA punishment was abandoned and they became eligible for the 2014 postseason, there was an immediate run of "Is Penn State the Big Ten's best Playoff hope?" -- they're the most well-positioned to win the East among teams not named Michigan State or Ohio State.

In the West, it's Wisconsin's race to lose. The Badgers rank as the division's best team, for starters, plus they get Nebraska at home. If the Huskers pull an upset in Madison on November 15, the projected wins for those two teams almost completely even out.

Sure bets

There's one thing to remember when dealing with odds instead of "I think Team A will beat Team B" types of predictions: they're conservative by nature.

If you have a 90 percent chance of beating your opponent, it means you're a lot better and will almost certainly win. It also means that you're only assigned 0.9 projected wins. Add up a lot of games like that, and almost nobody is projected to go undefeated. Hell, even now, Marshall, with a top-40 ranking and the easiest schedule known to man, is projected to only win 10.9 of 12 games this year.

Combine the general nature of odds with the fact that F/+ projections have been even more conservative (with 70 percent chances playing out like 80 percent chances, et al, so far), and you get the Big Ten's three superior teams (Michigan State, Ohio State, and Wisconsin) projected to win just 17.6 of 24 combined games.

I thought it would be interesting, then, to note how many "easy" games each team has. If we say that anything above 70 percent is a likely win, then we see that Wisconsin has six likely wins, Michigan State and Ohio State each have five, and only Penn State has more than two.

(On the flipside, Purdue has seven likely losses. Poor Purdue. You better beat Northwestern, Boilers.)

Team / Range 10-19%
Chance
20-29%
Chance
30-39%
Chance
40-49%
Chance
50-59%
Chance
60-69%
Chance
70-79%
Chance
80-89%
Chance
Illinois 3 1 2 1 1
Indiana 1 3 1 1 1 1
Iowa 2 2 2 2
Maryland 1 3 2 2
Michigan 2 1 3 2
Michigan State 3 3 2
Minnesota 1 3 1 1 1 1
Nebraska 2 1 1 2 2
Northwestern 1 2 3 2
Ohio State 1 1 1 4 1
Penn State (1-0) 2 2 2 1
Purdue 4 3 1
Rutgers (0-1) 1 2 2 1 1
Wisconsin 2 2 4

Indiana and Iowa both see six of eight games fall into the 30-70 percent range. Consider them major wildcards here, for better or worse, especially since both of Indiana's last two results (a loss to Bowling Green and a road win over Missouri) were quite unlikely. Odds may not apply to Kevin Wilson's Hoosiers this year.

Slog Level Delta

The "We ran 1,000/10,000/1,000,000 simulations!" thing might be a bit overdone by stat analysts these days, but it's the clearest way to explore what it means to have, say, 3.26 or 5.47 projected conference wins. I simulated the Big Ten's 55 remaining conference games 10,000 times to explore what the above probabilities mean for the conference race as a whole.

First, what record is it going to take to win each B1G division?

Record West Winner East Winner
4-4 0.1% 0.0%
5-3 7.3% 3.5%
6-2 36.1% 36.5%
7-1 40.9% 46.8%
8-0 15.6% 13.2%

Ignoring ties in the standings, this shows us that there's a seven percent chance the West winner is 5-3 or worse. There's also still about a three-in-five chance that each race produces a runaway, 8-0 or 7-1 winner (Wisconsin in the West, Michigan State or Ohio State in the East).

Odds of Big Ten conference records

This week

There are five conference games this week in the Big Ten:

  • Iowa at Purdue (12:00 p.m. ET, BTN). Win probability: Iowa 76 percent.
  • Northwestern at Penn State (12:00 p.m. ET, BTN). Win probability: Penn State 80 percent.
  • Maryland at Indiana (1:30 p.m. ET, BTN). Win probability: Indiana 68 percent.
  • Minnesota at Michigan (3:30 p.m. ET, ABC/ESPN2). Win probability: Michigan 78 percent.
  • Illinois at Nebraska (9:00 p.m. ET, BTN). Win probability: Nebraska 84 percent.

(There are also four non-conference games on the docket, headlined by Cincinnati-Ohio State, but we'll focus on these five.)

Hey, we finally joined Facebook!

Only one of this week's five conference games features a spread under 10 points, though "Michigan -12" seems a bit on the aggressive side, considering Michigan's current state. As you would expect, then, only one game has win probabilities anywhere close to even.

So who has the most to gain or lose this weekend? Probably Maryland and Indiana. A big upset like Illinois over Nebraska or Northwestern over Penn State would redefine how we look at everybody involved, but in the week's most tightly projected contest, bowl hopes (Maryland) or darkhorse Big Ten title hopes (Indiana) might hinge on a win.

Maryland went 3-1 in non-conference play, dropping a heartbreaker to West Virginia and handling JMU, USF, and Syracuse. The Terps need to go at least 3-5 to become bowl eligible for the second straight year, but the probabilities say there's about a 37 percent chance they don't reach that mark. Beat Indiana, and the odds fall to 18 percent.

The Hoosiers, on the other hand, could have a role to play in the East. The Hoosiers have a solid chance of finishing either with a winning record in conference (again, if odds can actually be applied to the Hoosiers this year), and if they beat Randy Esdall's Terrapins, the odds of finishing 5-3 or better increase from 31 percent to 39 percent. The East is Michigan State's and Ohio State's hostage until or unless someone loses a game they shouldn't, but Indiana could make itself part of the conversation with a win on Saturday.